Occam's razor is more a heuristic at the level of epistemology than actual science (the latter involves the experimental method, not logical posits like Occam's Razor). And of course it's statistical, not absolute.
Heck, even in everyday life we find cases where Occam's Razor is plainly wrong. Sometimes the explanation for an event is the more bizarre one rather than the simpler one.
“The aim of science is to seek the simplest explanation of complex facts. We are apt to fall into the error of thinking that the facts are simple because simplicity is the goal of our quest. The guiding motto in the life of every natural philosopher should be “Seek simplicity and distrust it.” – Alfred North Whitehead
No. If you believe in science as a practice or a way to explain things or "operate" things (like someone that switch on a light and that have some knowledge about electricity).
If you agree with what science explains (how it works thus the Occam's Razor) and disagree with what science doesn't explain (yet or never), you create a duality.
The duality can be eliminated with the Occam's Razor and then remains what is explained by the science (with its hypothesis).
It's not an opinion. It's just a rationale.
Everything that will not fit in that paradigm will probably be rejected sooner or later (by science).
As writing and science has a beginning in human history, we may discover something more powerful than science to discover the rest.
I think he does a pretty good job of explaining a theory in which that isn't the case - that
a) in order to predict (compute) "you" at at point, you have to actually "run" you at every time-step up until that point; this is specifically analogous to NP-completeness: result verification is identical to result generation; which could be meaningfully rephrased as "there's no shortcut to figuring out what you will do than going out and doing it."
b) There's a gap between non-determinism and randomness.
This reminds me of the basics of Kolmogorov complexity: the number of bytes to transmit a random string is the same as the size of the random string, except instead of "random" you have "non-deterministic". You are a non-random non-deterministic string; there is no shortcut in computing you than actually doing it... with the rest of the universe. (Which may mean there's no way to transmit you, either...?)
>If free will depends on an inability to predict future behavior, then it would follow from that free will somehow depends on our being unique: on it being impossible to copy us.
That's similar to the conclusion I came to, too.
That free will is basically the series of singular inevitable outputs of a computation (the "me-the-person-in-the-universe" computation) that are not compressible/reducible (they have to be run to come to the output).
And it's free not because it "freely" selects among several choices, but rather because it's uniquely personal -- as a person, one can ever only do/will the thing they end up doing/willing (tautologically: one's history is he/her -- our will defines us), but what that is can only be found out after the tact.
>There were some famous experiments in the 1980's, where someone would attach electrodes to someone's brain and would tell them that they could either press button 1 or 2. Something like 200ms before the person was conscious of making the decision of which button to press, certainly before they physically moved their finger, you could see the neurons spiking for that particular finger. So you can actually predict which button the person is going to press a fraction of a second before they're aware of having made a choice. This is the kind of thing that forces us to admit that some of our choices are less free than they feel to us---or at least, that whatever is determining these choices acts earlier in time than it seems to subjective awareness.
Similar to my comment above, I find the ideas drawn from those experiments (that there's no free will) wrong.
The fact that a choice/action can be known (from neural activity etc) before the choice becomes conscious to the subject, doesn't necessarily mean that the subject doesn't have free will. It can just mean that the subject has a whole beneath-the-surface thing beyond the conscious layer. It's not like the person's neurons belong to somebody else -- or that we only "live" by thinking.
> It can just mean that the subject has a whole beneath-the-surface thing beyond the conscious layer.
But doesn't that mean at some point you can't say effect A was caused by action B? There has to be some point where there are some number of possible outcomes but even knowing the entire state, you cannot predict which path will be taken and that any particular path can be taken reliably.
> There has to be some point where there are some number of possible outcomes but even knowing the entire state, you cannot predict which path will be taken and that any particular path can be taken reliably.
Why? It's not predictable whether any random Turing machine will halt, but whether that specific machine halts still has a deterministic, well-defined answer.
What I was trying to convey was that unpredictability and and determinism aren't mutually exclusive. The "reliable path taken" could depend on solving the Halting problem, for instance, which we can't do but perhaps our natural laws somehow can.
But even if that's immaterial, it's not clear that determinism has any relevance to free will at all. You, like many others, have simply assumed that the "free" aspect of free will implicitly means "freedom from antecedent causes", but the long debate over free will in philosophy casts this interpretation in serious doubt.
>whether that specific machine halts still has a deterministic, well-defined answer
Really? If there is literally no procedure that can tell the difference between "machines that will never halt" and "machines that will run longer than you can afford to wait", you're basically calling for a concept of "deterministic" where it's ok if a final determination never happens, and in some cases (which you can't identify), can't happen. That seems like it's counter to the spirit of calling something deterministic.
Just because there's no general algorithm to determine whether Turing machine X may halt, does not preclude the existence of specific Turing machines that can determine whether X will halt. This is part of the difference between truth and proof. Whether X will halt is true or false regardless of whether to use can prove it to be true or false in any given logic.
As far as proofs go, there are few elegant free will thought experiments, may it be from philosophical, spiritual or scientific approach/kind, that are actually all for some level of free will in this system we live in.
But anything that have people guess something, and then conclude if free will exist or doesn't, just fails bunch of working techniques we have to coming to conclusions from experiments.
Well, if you knew the entire state, you could predict which path will be taken.
What I said was that that the "agent" having the "free will" in a person can be larger than their conscious layer. Our brains are still ours (and define us) even if we don't consciously know everything they do or all information they keep at all times.
> Similar to my comment above, I find the ideas drawn from those experiments (that there's no free will) wrong.
Exactly. Pointing out that the mechanism by which we make a free choice is slower than our detectors and computer equipment doesn't prove anything. I've always found this interpretation bizarre.
That free will is basically the series of singular inevitable outputs of a computation (the "me-the-person-in-the-universe" computation) that are not compressible/reducible (they have to be run to come to the output).
That seems a bit like goalpost shifting to me. What is the common usage of the term free will? I argue it is sitting in a restaurant and being able to order either steak or fish and that this decisions in neither deterministic, i.e. the choice is not a function of the state of the universe prior to the decision, nor random, i.e. the choice is not independent of the state of the universe prior to the decision.
I also argue that there is no third option and one could just define the second option as the negation of the first. This obviously renders this definition of free will inconsistent and leaves us with the choice between determinism and randomness as defined above. Of course only unless someone can point out why this definition does not capture the common usage of the term free will or that there is an error in the definition of or the relation between the two options.
Note that this way of phrasing it avoids the problem of talking about a probability distribution which might be problematic because we might be unable to repeat the experiment or have to deal with other complicating factors like the probability distribution itself being a function of the state of the universe prior to the decision.
What you are suggesting is obviously not how I think the term free will is ordinarily used and how it should therefore be defined. Your definition would clearly fall into my definition of no free will and only additionally demand that there is no efficient procedure to evaluate the function determining the decision so that we are unable to predict decisions at least in practice.
This is certainly not an unacceptable definition and I can not tell whether I would say there is free will or not under that definition. How hard would evaluating the function have to be in order to call it free will? Just really impractical because it would require measuring the state of billions of billions of particles with very high accuracy? Or actually impossible due to something like the uncertainty principle or the no-cloning theorem?
But anyway, I argue that this definition, even if it might be useful, does not match what the term free will is usually supposed to mean.
>That seems a bit like goalpost shifting to me. What is the common usage of the term free will? I argue it is sitting in a restaurant and being able to order either steak or fish and that this decisions in neither deterministic, i.e. the choice is not a function of the state of the universe prior to the decision, nor random, i.e. the choice is not independent of the state of the universe prior to the decision.
I'd say that's a false approach, as it doesn't capture the actual thing we mean by "free will" -- even if it seems it does initially.
Neither dependent on the state of the universe nor random basically implies a soul. That is, an agent outside of the universe (that can still know part of the universe's history as it pertains to them and consider it when they take their decision but is not influenced by it directly to the decision). In the end, that could just be a random choice.
Now, the essential when we say free will is that the agent that decides a course is "me".
In other words, the ability to take different choices when repeating the experiment with the exact same parameters (state of the universe) is not important to free will. Neither is the ability to not be influenced by the state of the universe important. The only important aspect is "will" -- so it's not free as "whatever goes", but free as in "mine".
And it wouldn't be mine if, given my timespace-history as a person, I would take entirely different decisions given the same exact initial conditions. I'm me precisely because I take specifically a certain decision -- mine decision.
Using the term "free will" casually some might think that this means we can decide arbitrarily (e.g. if we could repeat the experiment with all other conditions absolutely the same, we could possibly see a different decision), but I think that's a wrong extension of the basic requirement: that the decision comes from the agent.
And the way that this is ensured in my version (and in line with both determinism and randomness in the universe) is that the agent IS the decision (the sum of their decisions) embodied.
Which is also symmetrical: given the same circumstances exactly (to the last particle being the same etc) we would make the same decision -- but on the same time, it's our history of passing through those exact circumstances and making those decisions that defines us (our "I").
What those going for the "arbitrary free choice" (possibility of different choice on same experiment) do is that they define the agent as someone outside the universe, e.g. dualism. But we are "me" and "you" inside the universe, and only to the extend that we have our exact personal histories.
I'd say that a false approach, and that it doesn't capture the actual thing we mean by "free will" even if it seems it does.
I obviously have to disagree and I want to stress again that I am talking about the way the term free will is commonly used, not about the way it should be used. Go and ask people. I argue that most people will say that they could have made either choice in the sense that repeated trials would yield differing choices and that making the choice is not merely an act of flipping a coin.
This of course turns out to be an impossibility under close inspection, as I argued before, and it therefore follows that free will in the common usage of the term does not exist. But I am pretty sure most people will nonetheless object if you argue that they could not possibly have made any other choice than the one they made besides maybe by flipping a coin.
Neither dependent on the state of the universe nor random basically implies a soul. That is, an agent outside of the universe (that can still know part of the universe's history as it pertains to them and consider it when they take their decision but is not influenced by it directly to the decision). In the end, that could just be a random choice.
Such a kind of soul would be part of the universe as far as I am concerned. It is something that can be acted upon by the normal universe to record facts and it can act on the normal universe by influencing decisions. I am unable to see how such a kind of soul would be significantly different from a halo of weakly interacting dark matter surrounding every body, gravity aside.
There are weak bidirectional interaction between the normal matter of your body and the dark matter of your halo, and they are so weak that they escape our current detection capabilities. We also do not currently understand and are unable to observe the internal dynamics of dark matter halos.
Is there any aspect to souls that this analogy does not capture? Well, souls could possibly exist outside of space and maybe even time, but that seems to be nothing more than a minor technical detail, at least in the context of free will. So I argue that souls of the kind you described are not adding anything to the picture, they are mostly equivalent to some kind of yet undiscovered physics.
The essential when we say free will is that it's "me" the agent that decides.
It is undoubtedly true, that it is an important part of free will, that the decision originates from the agent. This aspect is, I argue, captured in the term will, the desires of an agent.
Some might think that this means we can decide arbitrarily (e.g. if we could repeat the experiment with all other conditions absolutely the same, we could possibly see a different decision), but I think that's a wrong extension of the basic requirement: that the decision is uniquely on the agent.
This is, I argue, the question of what the term free means. As argued before, I think most people actually understand it in the sense that you consider a wrong extension. I also think that in your point of view free would be mostly redundant, just affirming what will essentially already conveys, that the decision originates from the agent, that the decision is free from external influences.
And the way that this is ensured (and in line with both determinism and randomness in the universe) is that the agent is the decision (the sum of their decisions) embodied. Which is also kind of obvious: given the same circumstances exactly (to the last particle being the same etc) we would make the same decision. But it's also our history of passing through those exact circumstances and taking those decisions we took that makes us "I".
Here I just want to note that this interpretation might have its very own problems, especially what constitutes external influence? What if steak is out? Or if the restaurant is on fire? What if I force you at gunpoint to order fish? What if I kidnapped you twenty years ago and o...
I also just conducted a N = 1 study to support my argument. The result was mixed. The test subject started out supporting your point of view with strong emphasize on personal preferences and the claim that several repetitions of the experiment, i.e. traveling back in time a few minutes after a choice between steak and fish was made, would always yield the same choice. The response changed when I asked for simply choosing between two arbitrary numbers, 5485151218 or 1584512548, stating that repeated experiments would yield each choice with equal probability.
So the subject generally put strong emphasize on his personal history and preferences as well as on the circumstances and consequences of the decision. But if those considerations were eliminate by asking to make inconsequential decisions where the subject could hardly have any preferences, then the aspect of being able to make either choice became visible.
This made me better unterstand where your point of view comes from. In hindsight I have to admit that I unwittingly always heavily focused on inconsequential decisions without preferences. But I would now extend or change my argument and argue that this actually the right thing to do because those are the more interesting cases.
There were several options and I freely choose the one that seemed best to me. The use of freely and choose seems almost a bit paradoxically to me in this case. On the one hand it emphasizes the absence of constraints, probably with most emphasize on external constraints, and the availability of several options, on the other hand it implies that considerations of personal preferences, circumstances, and consequences naturally led to one choice.
So in what sense was the decision free if the choice is dictated by constraints? I can only think of partitioning possible constraints into two groups, those that someone wants to obey and those that one does not want to obey. You want to eat your favorite food but you don't want to be forced to eat some food.
This probably also has ramifications for how one has to look at different options. In what sense are there actually several options if the choice is dictated by constraints? For differing sets of options the choice may differ, but for a fixed set of available options there is only one choice. So it seems to me that the actual choice is between differing sets of available options, not between different options once the set of available options is fixed. But this choice is not a real choice either, the set of available options is fixed with the scenario the agent finds itself in.
Leaving the case behind, were the decisions are heavily dictated by constraint, I will finally return to the case with actual choice. This seems to be the much more interesting case, what if decisions are not dictated by constraints? Could you have chosen otherwise? I still argue that many if not most people would respond yes, of course, I could have done something else. Even if this position is hidden in many scenarios by considerations of constraints that dictate or at least seem to dictate one choice. I could have chosen differently but I would never have done so given the circumstances.
>I obviously have to disagree and I want to stress again that I am talking about the way the term free will is commonly used, not about the way it should be used. Go and ask people.
The "way most people understand it" is a little moot point, because most people don't really think about it that much. So when asked they just say some hazy and generic version of what it means.
Unlike programmers and logicians, lay people are quite bad at giving precise definitions -- and thinking what those definitions imply.
Philosophy has investigated free will for millennia, and has frequently (if not almost always) gone beyond the naive immediate understanding of the term by laymen.
>Here I just want to note that this interpretation might have its very own problems, especially what constitutes external influence? What if steak is out? Or if the restaurant is on fire? What if I force you at gunpoint to order fish? What if I kidnapped you twenty years ago and only gave you fish for four years so that you hate fish today? What if you are under the influence of drugs?
Those are still fine for free will. Even at gunpoint, one is still the one taking the choice (they could chose to die just as well). Free will doesn't imply arbitrarily bending the universe to one's will, or that nobody puts hard problems on the agent of free will.
Else, Sophie wouldn't have much reasons to feel bad for her choice.
>and I would not be to surprised if, upon close inspection and careful definition, it turns out that the agent is identical with the past light cone of the decision.
Perhaps, but the decisions are made at the edge of different light cones -- which are the different persons.
>Unfortunately, I was elected to a tutorial fellowship at Merton, my old college in Oxford, and was submerged in the pressures of Oxford tutorial life, so that my Freedom of the Will was not published until 1970. It is not very different from the original article, but does meet some of the criticisms first leveled against it, and sets the Godelian argument in a more general context. Perhaps two points I make in the book are worth reiterating. Professor Minsky spent some time yesterday criticizing Professor Margenau saying that he had misunderstood the import of quantum mechanics. I think actually the misunderstanding was on Professor Minsky's part and that he had failed to understand the structure of Professor Margenau's argument. Quantum mechanics is relevant to the problem of the will because it has replaced classical Newtonian physics which seemed to rule it out. It does not of itself prove free will - lots of quantum mechanical systems have no free will - but it disproves a disproof of it. In my book I referred to speculations on whether quantum mechanics might be replaced by a more determinist theory - a "Hidden Variable Theory" as it is called, and Von Neumann's argument against it. Since then the argument has progressed much further. Von Neumann's proof has been proved and criticized, and a whole series of results have been obtained - Bell's inequalities, Gleason's Theorem, Kochen-Specher Theorem, the two-color theorem: and four years ago Aspect concluded some experiments in Paris, which seemed to rule out any prospect of a hidden variable theory. So it seems to me that Professor Margenau was quite right to see quantum mechanics as bearing on the freewill problem, not as proving that it exists but as, in Plantinga's terminology, defeating a defeater.
further, Lucas proposed the following:
1. Determinism ↔ For any human h there exists at least one (deterministic) logical system L(h) which reliably predicts h's actions in all circumstances.
2. For any logical system L a sufficiently skilled mathematical logician (equipped with a sufficiently powerful computer if necessary) can construct some statements T(L) which are true but unprovable in L. (This follows from Gödel's first theorem.)
3. If a human m is a sufficiently skilful mathematical logician (equipped with a sufficiently powerful computer if necessary) then if m is given L(m), he or she can construct T(L(m)) and
Determine that they are true—which L(m) cannot do.
4. Hence L(m) does not reliably predict m's actions in all circumstances.
5. Hence m has free will.
6. It is implausible that the qualitative difference between mathematical logicians and the rest of the population is such that the former have free will and the latter do not.
It seems to me that step 3 requires that the human m have a sufficiently accurate understanding of themselves. That step seems to me to be questionable.
he's arguing a proof of mere existence there. You don't need to propose an algorithm, or indeed have any idea how to go about it, you can just assert that it must exist and that's enough.
A physicist, an engineer, and a mathematician are in a hypothetical theater when the stage curtains catch fire.
The physicist makes some measurements as to the volume of curtain, available air flow, chemical properties of the curtain, flow rate and thermal capacity of the water hose and water, formulating a precise equation for how much water is necessary to put the fire out when starting at a given time, and precisely puts the fire out.
The engineer makes a few quick estimates as the the volume of fuel and air, uses that to estimate the amount of water needed, doubles it, and puts the fire out (with safety factor 2).
The mathematician looks at the fire, looks at the fire hose, states: "The problem is solvable", and exits stage left.
And I'm arguing that his proof of mere existence has holes in it. At a minimum, step 3 is not a mechanistic step that can be simply asserted.
Or, rather, I assert that step 3 posits a degree of self-knowledge that no human actually possesses. Now you have to choose which unproven assertion you're going to believe, his or mine, or else you have to find a way to prove one or the other.
> Bell's inequalities, Gleason's Theorem, Kochen-Specher Theorem, the two-color theorem: and four years ago Aspect concluded some experiments in Paris, which seemed to rule out any prospect of a hidden variable theory. So it seems to me that Professor Margenau was quite right to see quantum mechanics as bearing on the freewill problem, not as proving that it exists but as, in Plantinga's terminology, defeating a defeater.
It seems he was ignorant of Bohmian mechanics then, because deterministic QM is still alive and well.
> 3. If a human m is a sufficiently skilful mathematical logician (equipped with a sufficiently powerful computer if necessary) then if m is given L(m), he or she can construct T(L(m)) and Determine that they are true—which L(m) cannot do.
This step is dubious:
1. It's not at all clear that m could prove the truth of T(L(m)) even in these circumstances. I would need to see these steps to be convinced.
2. Humans are finite automotons (see the Bekenstein Bound), and the addition of a "sufficiently powerful computer" drastically increases the computational power available to m beyond what m is capable of alone. So L(m) would have to be a logical system that encompasses this unbounded power, and so T(L(m)) would still have true but unprovable statements, even with the computer due to Goedel.
This argument basically tries to prove that a Turing machine + Oracle can solve the Halting problem for Turing machines, but cannot solve the Halting problem for Turing machines+Oracles. However, it neglects to acknowledge that L is predicated on the Oracle, so we're not solving the Halting problem for Turing machines, we're solving the Halting problem for Turing machines+Oracles given only an Oracle, which is impossible.
As for the argument itself, it's a clever attempt, but still predicated on the assumption that determinism is inconsistent with free will. This has yet to be demonstrated, and in fact, all such arguments to date have been refuted (although clearly not to everyone's satisfaction).
In my limited understand of philosophy, it seems this entire lecture can be boiled down to one question (which I didn't think was answered). Does knowledge of a decision preclude free will?
The core of my argument is going to be agency. We are in control because our actions often arise out of our own agency. This as opposed to there being another agent being able to decide our actions.
For this argument to work at all, we need to define agency/being an agent and establish that humans are agents. Moreover, humans must be at least somewhat unique for having agency. Certainly, we don't want rocks, or calculators to be ascribed agency.
I'd say an agent needs 2 things:
- Intention
- 'Potency', that is the posibility to measure and affect things related to those intentiosn.
This has again moved the goalposts, since we now need to establish Intention and potency. I'd posit that, if humans have intention, they certainly have Potency and thus only focus on intention.
Intention arose in humans through evolution. Skipping some steps, at one point organisms started measuring their environment, and responding to it. This is a huge evolutionary advantage (i.e. move towards the food and away from the predators). Simple measurement and response isn't quite enough though, but there is another advantage to be had: modeling. An organism is modeling when it isn't just measuring the outside world, but also making predictions. Once these predictions start taking into account possible actions by that organism, and those predictions start informing the actions of the organism, those actions have 'intent'.
And that is my argument in short form. Free will arises from choosing actions based on their predicted effect in some internal model.
There is some stuff to clear up between being an agent, and having agency over your own actions. It might be some other much more effective agent has more control over you than your own agency. I don't believe any such powerful agents exist though.
Neither the calculator nor the bad guy has intention. They lack a feed-back-loop between potential decision and internal model that evaluates that decision.
For the videogame UI, there is a massive limitation on agency because it was fully designed by another agent. That designer agent thus has very large influence over the AI.
I'd allow for complicated AI with emergent behavior to have agency up to a point. As a corollary, this means humans could have agency even if they were created by god.
I think this is pretty easy. We as people have likes, dislikes (i.e. preferred states), desires, etc. We also identify with these preferred states and desires. Furthermore, we have the capacity to alter the world and our behavior to bring about these preferred states and desires. To put it another way, we can alter our behavior for the sake of achieving desired goals.
This is a higher level of self control than a calculator in that it has no capacity to alter its behavior. If a calculator had preferred states, it has no capacity to alter its behavior to achieve those preferred states.
Now, we still don't have the capacity to change our will and so we don't have "ultimate" (contra-causal) free will. But I think its pretty clear that we have more control than a calculator.
I'd like to reword what I meant: we do technically have more control than a calculator. I guess what I mean is something more like: how are we any less deterministic than a calculator?
I don't mean, harder to predict. I mean, are we not as deterministic as anything else?
Sorry meant to reply to this sooner. Yeah, we're just as deterministic. The question then is, does determinism invalidate choice? What is it about choice that necessarily makes it contradict determinism? Is it the idea that the choice is predictable under determinism? Is it that there was ultimately only one option available for you and thus only the illusion of choice? If that's the case then we don't have that kind of choice.
The way i see it is that choice/free will is just a useful concept at the level of description of people. The mistake is mixing the levels of description and expecting the resulting statements to be coherent. There's no point in a causal chain where a "choice" happens, yet at the level of description of people with behaviors, choice happens. If we don't expect some kind of metaphysical truth underlying the claim that we make choices, then there's no problem. If you want your choices to have some acausal sort of control, then you're out of luck.
Ugh, another non-philosopher hacking at straw men with poorly defined terms. The equivalence of non-determinism and randomness is not at all necessary to dispel the idea of free will; simply ask someone to define it, and it either a) turns into vapor or b) turns to abstraction that is equally meaningless, such as to knowledge. What is knowledge but state? It has nothing to do with free will, which cannot be differentiated from any other physical phenomenon.
It's also a terribly boring problem. Regardless on you stance on what free will is, the answer does not mean you are removed from culpability; it rather means that you are not exempt from defense of your actions as a being with finite knowledge explaining your state to another being (presumably, a court) with finite knowledge.
For all the philosophers asking this question, they fail to produce any instance where the answer is meaningful. Leopold and Loeb managed to persuade a gullible jury that they can excuse human interactions with philosophy; it is of interest to pop psychology and flaws in the american justice system, NOT to philosophy.
The phenomena of consciousness are far more worthwhile concepts because they don't pigeonhole reality into a dry dichotomy.
AFAICT, his terms are very well defined; he's using terms specifically from quantum computing, which combines terms from quantum phyics, classical computing theory, and adds a few of its own.
He's also doing the thing where he's discussing the issue without a direct attempt to resolve it, which may be throwing off your comprehension.
> It's also a terribly boring problem. Regardless on you stance on what free will is, the answer does not mean you are removed from culpability
If you acknowledge culpability in a given set of circumstances, then those culpable acted of their own free will. Choice from free will entails moral culpability, so you can only deny free will by also denying moral culpability.
The fact that it's difficult to define the precise meaning of these terms doesn't make the problem meaningless.
Culpability is a happy illusion outside of a social context. Why are we trying to define this in quantum physics terms? It only makes sense in a Kantian universe, where there is an ideal of rationalism. Given how ambiguous human language (let alone human thought) is, do you seriously think that there is an objective culpability for arbitrary concepts when people won't even interpret the arbitrary concepts the same way?
All of this smacks of scientists trying to weigh in on ontological problems. Science is only relevant so far as to show us the limits of what we can claim—in this case, science shows us free will doesn't map neatly into our known universe, hence the article above trying to argue for ways to preserve this abstract concept in a world that doesn't care.
Look at: The Emperor's New Mind. It's the best attempt so far at actually fleshing out the problem, and it hinges entirely on the (fallacious) idea that because humans can innovate they aren't bound by turing completeness. Of course, this is hogwash, because what you'd have to demonstrate was that a single consciousness can prove everything given a finite set of axioms, which is of course an impossible proposition. Penrose sidesteps this by focusing on the magnification of quantum effects to human-level.
I mean, come on—if that's your hope at understanding your own brain, just give up and imagine fairies in it. Is that really more comforting (...let alone rational) than accepting you're driven more by your circumstances than by some ineffable thing about yourself you can't even define? Might as well accept that you're inherently an observer to phenomena outside your immediate ability to understand completely—you know, the basis of the scientific theory.
When people talk about free will in terms of quantum effects, all they reveal is how poorly read they are. It's a binary concept that doesn't map at all into a world where things-are—-—of-predictatable-but-you-can't-predict-the-unpredictable. It's nonsensical except at levels that are meaningless to human thought OR meaningless to physics.
> Culpability is a happy illusion outside of a social context
I'm sure that's what you believe, but that doesn't make it true.
> do you seriously think that there is an objective culpability for arbitrary concepts when people won't even interpret the arbitrary concepts the same way?
The whole point of objectivity is that interpretation has no bearing on truth (or culpability in this case).
> in this case, science shows us free will doesn't map neatly into our known universe
Science shows no such thing. Assuming determinism of choice shoes only that philosophies of free will that claim free choice is incompatible with determinism are incompatible with evidence. Of course, we don't actually know that we are deterministic (evidence is not conclusive), and it does not in any way rule out free will that is compatible with determinism, which is the type most philosophers profess.
This is an all too common mistake that scientists make: thinking that their definition of free will is obviously the one everyone is using.
There is a part in this documentary that covers a man who had a hypothesis that human beings are driven to do things by their DNA. I'd link the specific scene but I have to go to work, maybe later.
Speaking of Bell's inequality, according to Erik Verlinde's theory on emergent gravity, gravity is an emergent force from the supersymmetry relationship of fundamental particles across the entirety of space and time. Each particle was encoded with a quantum memory state before a thermal energy source disrupted the tiny space and separated the paired particles across the galaxy. If this is true, I believe the observed violation of Bell's inequality could be explained by this pre space-time deterministic relationship that causes the force of gravity to emerge.
I'm sad that this article is on the front page of HN. It's clever in a few places, but makes basic errors. In short, it's not a competent presentation of the topic.
What he calls the "free will camp" sounds like a weird mishmash of libertarianism (determinism is false, and we have free will) and hard-determinism (determinism is true, therefore we don't have free will). It completely ignores compatibilism (https://plato.stanford.edu/entries/compatibilism/).
Moreover, the argument he gives regarding Leopold and Loeb is baffling. That a jury has no free will to listen to your argument for hard-determinism is prima-favor irrelevant. A hard-determinist holds that some people are determined to believe in free will, and moral responsibility, but they're incorrect. Any graduate student in philosophy would be expected to note these distinctions.
As for the view that free will is an illusion, he presents one of the canonical arguments uncharitably. Leaving aside the question of whether the invocation of NP actually establishes what he says it does (something non-determined but non-random), the opponent of libertarianism doesn't need the claim that determinism and randomness are our only options. They just need (something like) the premise that if our actions don't follow from our reasons/emotions/volition, they can't be considered free, and that anyone proposing some status between randomness and determinism needs to account for the sense in which decisions flowing from that status are free.
The linked lecture is a solid, engaging and informative discussion of an issue, bringing together many different fields of though, science, and engineering. That sounds like a great thing to have on the front page.
Hmm. I can see that. It referenced (without much explanation) many other ideas, and provided amazing and novel combinations of those ideas, but if you wanted "informative" you'd need to google the parts you don't know.
Edit: That is, it "taught" me little outside of amazing and novel new perspectives on things I already know.
It's probably in between. It's incredible how he does not talk about compatibilism in a serious post about different possibilities of free will. I don't understand how that can be missed on purpose. But, his arguments and content were coherent and interesting. I don't think it hurts HN to have this on the front page.
Can you explain compatabilism, particularly how you see it relating to what he's talking about? The wikipedia explanation is... underwhelming, and the next best thing I saw was practically a textbook.
Near as I can tell, from the Wikipedia explanation, compatabilism is an ethical consideration of free will. Most other philosophy is on a metaphysical level; Scott Aaronson appears to be talking about free will as a physical phenomenon. It make sense that he would't bring up an ethical framework.
Have you seen the Stanford Encyclopedia of Philosophy? Some articles are too technical to read without a background, but many are good for anyone: https://plato.stanford.edu/entries/compatibilism/.
Here's my take, which is slightly different in emphasis than the SEP one (which I guess means I am probably wrong/idiosyncratic):
Compatibilism is a metaphysical theory. What is counter-intuitive about the free will debate is that there are two entwined but distinct questions.
The first is (roughly) what is the full physical/causal/predictive/descriptive* story about human decision making. Is it deterministic? Almost deterministic but with some random noise? Something else?
Second, given that description, do we have free will? That is a metaphysical question, though many people are interested because they believe that there are ethical implications (compare: whether God exists is a metaphysical or perhaps empirical question, but it might have some ethical implications).
Compatibilism is primarily an answer to the second question. It says roughly "you might have thought free will means not being determined, but that is an illusion. Actually, free will is completely compatible with determinism. It is something like being able to act on the things that you take to be reasons, even if it's fully deterministic which things you think are reasons."
Talking about free will as a physical phenomenon (which is not quite what I'd describe Aaronson as doing) typically indicates that you've assumed some answer to the second question, so that you're only considering the first.
Btw: one point of view is that the first question makes sense, but the second doesn't. I don't have a developed opinion, as I didn't really work on this topic when I studied philosophy, but I lean that way.
* (none of these terms is perfect, and there's lots of debate about the right characterization here)
This is a really hard question without knowing what your area of interest is.
However, I find Descartes Meditations gripping for the way he poses the question of whether there can be any solid foundation for knowledge and the resulting skeptical worries (though his solution is somewhat disappointing). Hume's Enquiry Concerning Human Understanding is also wonderful.
If you find free will interesting, Kane's "A Contemporary Introduction to Free Will" is solid--it was the text a professor assigned when I ta'd for his course.
I'm very interested in math, and would say I'm fairly comfortable with abstraction (e.g. categorical thinking). I don't find philosophy of math or foundations very engaging, though.
I think the biggest answer that's in here to The Question is this one:
> For me, the glaring fallacy in the argument lies in the implication Not Determined ⇒ Random. If that was correct, then we couldn't have complexity classes like NP---we could only have BPP. The word "random" means something specific: it means you have a probability distribution over the possible choices. In computer science, we're able to talk perfectly coherently about things that are non-deterministic, but not random.
...which, in a way I don't fully grok, seems to be saying there's room between things being predicable (no free will) and random (nonsensical).
Then later, this bit:
> 1) We have the free will to choose in what basis to measure a quantum state. That is, at least the detector settings are not predetermined by the history of the universe.
2) Relativity gives some way for two actors (Alice and Bob) to perform a measurement such that in one reference frame Alice measures first, and in another frame Bob measures first.
3) The universe cannot coordinate the measurement outcomes by sending information faster than light.
which I also still don't really understand but woah.
As a contribution: I encountered an amazing demonstration of "free will", in that it was an experience you could go through, draw a box around, and write "here be free will" on the box.
It's pretty simple: Think of a city, and say it's name aloud.
Now consider what just happened in your mind and brain.
You didn't say the name of a city you've never heard of. You could not do this; you were "not fee" to make that "choice".
You (probably?) though of more than one city, and then selected on. You could do this; you were "free" to make that "choice".
This demonstration doesn't really go after the question as to what you actually have that can be labeled "choice", but it really nicely draws a box for that label to be applied to.
The way I understand the non-determined != random argument is like this: you look to your left and will or will not find a briefcase with a million dollars. It is not determined that you will not find said briefcase, since there is a probability, no matter how small, that it has materialized there in some way. However, the presence of the briefcase is not random, because there is not an even probability of it being there or not. Randomness means equal probabilities of the outcomes.
Hmmm I don't think that's it. It sounds more like the "randomness" of Go, microservice architectures, and the butterfly effect, except based in quantum interaction theory. That is, the overall complexity is such that the outcome cannot be determined (or is very hard to determine) prior to the outcome actually occurring; and then the quantum/relativity stuff makes that more than just about available computation power.
Lately I tend to think, that talking about “free will“ makes no sense at all. Our consciousness is the product of physical processes. They lead to the decisions we make. That's all there is.
Now if you bring a free will into play, where does it reside? Is it somewhere outside physics? Thats metaphysics then. And out of reach for any understanding anyway so pointless to reason about.
What I could agree on is to see “free will“ as a metaphor for the essence of our existence. The part that we probably never can explain.
80 comments
[ 0.22 ms ] story [ 135 ms ] threadOccam's razor is more a heuristic at the level of epistemology than actual science (the latter involves the experimental method, not logical posits like Occam's Razor). And of course it's statistical, not absolute.
Heck, even in everyday life we find cases where Occam's Razor is plainly wrong. Sometimes the explanation for an event is the more bizarre one rather than the simpler one.
“The aim of science is to seek the simplest explanation of complex facts. We are apt to fall into the error of thinking that the facts are simple because simplicity is the goal of our quest. The guiding motto in the life of every natural philosopher should be “Seek simplicity and distrust it.” – Alfred North Whitehead
http://scienceblogs.com/developingintelligence/2007/05/14/wh...
No. If you believe in science as a practice or a way to explain things or "operate" things (like someone that switch on a light and that have some knowledge about electricity).
If you agree with what science explains (how it works thus the Occam's Razor) and disagree with what science doesn't explain (yet or never), you create a duality.
The duality can be eliminated with the Occam's Razor and then remains what is explained by the science (with its hypothesis).
It's not an opinion. It's just a rationale.
Everything that will not fit in that paradigm will probably be rejected sooner or later (by science).
As writing and science has a beginning in human history, we may discover something more powerful than science to discover the rest.
Moreover, is the non-existence of free will really the simpler alternative to the existence of free will?
I would say yes because it's mostly a western world concern (thus cultural).
It was also used to explain the existence of the sin.
https://en.wikipedia.org/wiki/Thomas_Aquinas
a) in order to predict (compute) "you" at at point, you have to actually "run" you at every time-step up until that point; this is specifically analogous to NP-completeness: result verification is identical to result generation; which could be meaningfully rephrased as "there's no shortcut to figuring out what you will do than going out and doing it."
b) There's a gap between non-determinism and randomness.
This reminds me of the basics of Kolmogorov complexity: the number of bytes to transmit a random string is the same as the size of the random string, except instead of "random" you have "non-deterministic". You are a non-random non-deterministic string; there is no shortcut in computing you than actually doing it... with the rest of the universe. (Which may mean there's no way to transmit you, either...?)
Also, it's transcription of a lecture, which is much less common than YouTube videos of them, but a hundred times better IMHO
That's similar to the conclusion I came to, too.
That free will is basically the series of singular inevitable outputs of a computation (the "me-the-person-in-the-universe" computation) that are not compressible/reducible (they have to be run to come to the output).
And it's free not because it "freely" selects among several choices, but rather because it's uniquely personal -- as a person, one can ever only do/will the thing they end up doing/willing (tautologically: one's history is he/her -- our will defines us), but what that is can only be found out after the tact.
>There were some famous experiments in the 1980's, where someone would attach electrodes to someone's brain and would tell them that they could either press button 1 or 2. Something like 200ms before the person was conscious of making the decision of which button to press, certainly before they physically moved their finger, you could see the neurons spiking for that particular finger. So you can actually predict which button the person is going to press a fraction of a second before they're aware of having made a choice. This is the kind of thing that forces us to admit that some of our choices are less free than they feel to us---or at least, that whatever is determining these choices acts earlier in time than it seems to subjective awareness.
Similar to my comment above, I find the ideas drawn from those experiments (that there's no free will) wrong.
The fact that a choice/action can be known (from neural activity etc) before the choice becomes conscious to the subject, doesn't necessarily mean that the subject doesn't have free will. It can just mean that the subject has a whole beneath-the-surface thing beyond the conscious layer. It's not like the person's neurons belong to somebody else -- or that we only "live" by thinking.
But doesn't that mean at some point you can't say effect A was caused by action B? There has to be some point where there are some number of possible outcomes but even knowing the entire state, you cannot predict which path will be taken and that any particular path can be taken reliably.
Why? It's not predictable whether any random Turing machine will halt, but whether that specific machine halts still has a deterministic, well-defined answer.
But even if that's immaterial, it's not clear that determinism has any relevance to free will at all. You, like many others, have simply assumed that the "free" aspect of free will implicitly means "freedom from antecedent causes", but the long debate over free will in philosophy casts this interpretation in serious doubt.
Really? If there is literally no procedure that can tell the difference between "machines that will never halt" and "machines that will run longer than you can afford to wait", you're basically calling for a concept of "deterministic" where it's ok if a final determination never happens, and in some cases (which you can't identify), can't happen. That seems like it's counter to the spirit of calling something deterministic.
Add to that muscle charging time.
Add to that tissue acceleration.
Etc.
The fact that the brain decided 200 [ms] before the finger moved is just plain newtonian physics in the macro world.
But, modern journalism just have to hype it.
https://www.theguardian.com/science/2015/aug/27/study-delive...
As far as proofs go, there are few elegant free will thought experiments, may it be from philosophical, spiritual or scientific approach/kind, that are actually all for some level of free will in this system we live in.
But anything that have people guess something, and then conclude if free will exist or doesn't, just fails bunch of working techniques we have to coming to conclusions from experiments.
What I said was that that the "agent" having the "free will" in a person can be larger than their conscious layer. Our brains are still ours (and define us) even if we don't consciously know everything they do or all information they keep at all times.
Exactly. Pointing out that the mechanism by which we make a free choice is slower than our detectors and computer equipment doesn't prove anything. I've always found this interpretation bizarre.
That seems a bit like goalpost shifting to me. What is the common usage of the term free will? I argue it is sitting in a restaurant and being able to order either steak or fish and that this decisions in neither deterministic, i.e. the choice is not a function of the state of the universe prior to the decision, nor random, i.e. the choice is not independent of the state of the universe prior to the decision.
I also argue that there is no third option and one could just define the second option as the negation of the first. This obviously renders this definition of free will inconsistent and leaves us with the choice between determinism and randomness as defined above. Of course only unless someone can point out why this definition does not capture the common usage of the term free will or that there is an error in the definition of or the relation between the two options.
Note that this way of phrasing it avoids the problem of talking about a probability distribution which might be problematic because we might be unable to repeat the experiment or have to deal with other complicating factors like the probability distribution itself being a function of the state of the universe prior to the decision.
What you are suggesting is obviously not how I think the term free will is ordinarily used and how it should therefore be defined. Your definition would clearly fall into my definition of no free will and only additionally demand that there is no efficient procedure to evaluate the function determining the decision so that we are unable to predict decisions at least in practice.
This is certainly not an unacceptable definition and I can not tell whether I would say there is free will or not under that definition. How hard would evaluating the function have to be in order to call it free will? Just really impractical because it would require measuring the state of billions of billions of particles with very high accuracy? Or actually impossible due to something like the uncertainty principle or the no-cloning theorem?
But anyway, I argue that this definition, even if it might be useful, does not match what the term free will is usually supposed to mean.
I'd say that's a false approach, as it doesn't capture the actual thing we mean by "free will" -- even if it seems it does initially.
Neither dependent on the state of the universe nor random basically implies a soul. That is, an agent outside of the universe (that can still know part of the universe's history as it pertains to them and consider it when they take their decision but is not influenced by it directly to the decision). In the end, that could just be a random choice.
Now, the essential when we say free will is that the agent that decides a course is "me".
In other words, the ability to take different choices when repeating the experiment with the exact same parameters (state of the universe) is not important to free will. Neither is the ability to not be influenced by the state of the universe important. The only important aspect is "will" -- so it's not free as "whatever goes", but free as in "mine".
And it wouldn't be mine if, given my timespace-history as a person, I would take entirely different decisions given the same exact initial conditions. I'm me precisely because I take specifically a certain decision -- mine decision.
Using the term "free will" casually some might think that this means we can decide arbitrarily (e.g. if we could repeat the experiment with all other conditions absolutely the same, we could possibly see a different decision), but I think that's a wrong extension of the basic requirement: that the decision comes from the agent.
And the way that this is ensured in my version (and in line with both determinism and randomness in the universe) is that the agent IS the decision (the sum of their decisions) embodied.
Which is also symmetrical: given the same circumstances exactly (to the last particle being the same etc) we would make the same decision -- but on the same time, it's our history of passing through those exact circumstances and making those decisions that defines us (our "I").
What those going for the "arbitrary free choice" (possibility of different choice on same experiment) do is that they define the agent as someone outside the universe, e.g. dualism. But we are "me" and "you" inside the universe, and only to the extend that we have our exact personal histories.
I obviously have to disagree and I want to stress again that I am talking about the way the term free will is commonly used, not about the way it should be used. Go and ask people. I argue that most people will say that they could have made either choice in the sense that repeated trials would yield differing choices and that making the choice is not merely an act of flipping a coin.
This of course turns out to be an impossibility under close inspection, as I argued before, and it therefore follows that free will in the common usage of the term does not exist. But I am pretty sure most people will nonetheless object if you argue that they could not possibly have made any other choice than the one they made besides maybe by flipping a coin.
Neither dependent on the state of the universe nor random basically implies a soul. That is, an agent outside of the universe (that can still know part of the universe's history as it pertains to them and consider it when they take their decision but is not influenced by it directly to the decision). In the end, that could just be a random choice.
Such a kind of soul would be part of the universe as far as I am concerned. It is something that can be acted upon by the normal universe to record facts and it can act on the normal universe by influencing decisions. I am unable to see how such a kind of soul would be significantly different from a halo of weakly interacting dark matter surrounding every body, gravity aside.
There are weak bidirectional interaction between the normal matter of your body and the dark matter of your halo, and they are so weak that they escape our current detection capabilities. We also do not currently understand and are unable to observe the internal dynamics of dark matter halos.
Is there any aspect to souls that this analogy does not capture? Well, souls could possibly exist outside of space and maybe even time, but that seems to be nothing more than a minor technical detail, at least in the context of free will. So I argue that souls of the kind you described are not adding anything to the picture, they are mostly equivalent to some kind of yet undiscovered physics.
The essential when we say free will is that it's "me" the agent that decides.
It is undoubtedly true, that it is an important part of free will, that the decision originates from the agent. This aspect is, I argue, captured in the term will, the desires of an agent.
Some might think that this means we can decide arbitrarily (e.g. if we could repeat the experiment with all other conditions absolutely the same, we could possibly see a different decision), but I think that's a wrong extension of the basic requirement: that the decision is uniquely on the agent.
This is, I argue, the question of what the term free means. As argued before, I think most people actually understand it in the sense that you consider a wrong extension. I also think that in your point of view free would be mostly redundant, just affirming what will essentially already conveys, that the decision originates from the agent, that the decision is free from external influences.
And the way that this is ensured (and in line with both determinism and randomness in the universe) is that the agent is the decision (the sum of their decisions) embodied. Which is also kind of obvious: given the same circumstances exactly (to the last particle being the same etc) we would make the same decision. But it's also our history of passing through those exact circumstances and taking those decisions we took that makes us "I".
Here I just want to note that this interpretation might have its very own problems, especially what constitutes external influence? What if steak is out? Or if the restaurant is on fire? What if I force you at gunpoint to order fish? What if I kidnapped you twenty years ago and o...
So the subject generally put strong emphasize on his personal history and preferences as well as on the circumstances and consequences of the decision. But if those considerations were eliminate by asking to make inconsequential decisions where the subject could hardly have any preferences, then the aspect of being able to make either choice became visible.
This made me better unterstand where your point of view comes from. In hindsight I have to admit that I unwittingly always heavily focused on inconsequential decisions without preferences. But I would now extend or change my argument and argue that this actually the right thing to do because those are the more interesting cases.
There were several options and I freely choose the one that seemed best to me. The use of freely and choose seems almost a bit paradoxically to me in this case. On the one hand it emphasizes the absence of constraints, probably with most emphasize on external constraints, and the availability of several options, on the other hand it implies that considerations of personal preferences, circumstances, and consequences naturally led to one choice.
So in what sense was the decision free if the choice is dictated by constraints? I can only think of partitioning possible constraints into two groups, those that someone wants to obey and those that one does not want to obey. You want to eat your favorite food but you don't want to be forced to eat some food.
This probably also has ramifications for how one has to look at different options. In what sense are there actually several options if the choice is dictated by constraints? For differing sets of options the choice may differ, but for a fixed set of available options there is only one choice. So it seems to me that the actual choice is between differing sets of available options, not between different options once the set of available options is fixed. But this choice is not a real choice either, the set of available options is fixed with the scenario the agent finds itself in.
Leaving the case behind, were the decisions are heavily dictated by constraint, I will finally return to the case with actual choice. This seems to be the much more interesting case, what if decisions are not dictated by constraints? Could you have chosen otherwise? I still argue that many if not most people would respond yes, of course, I could have done something else. Even if this position is hidden in many scenarios by considerations of constraints that dictate or at least seem to dictate one choice. I could have chosen differently but I would never have done so given the circumstances.
The "way most people understand it" is a little moot point, because most people don't really think about it that much. So when asked they just say some hazy and generic version of what it means.
Unlike programmers and logicians, lay people are quite bad at giving precise definitions -- and thinking what those definitions imply.
Philosophy has investigated free will for millennia, and has frequently (if not almost always) gone beyond the naive immediate understanding of the term by laymen.
>Here I just want to note that this interpretation might have its very own problems, especially what constitutes external influence? What if steak is out? Or if the restaurant is on fire? What if I force you at gunpoint to order fish? What if I kidnapped you twenty years ago and only gave you fish for four years so that you hate fish today? What if you are under the influence of drugs?
Those are still fine for free will. Even at gunpoint, one is still the one taking the choice (they could chose to die just as well). Free will doesn't imply arbitrarily bending the universe to one's will, or that nobody puts hard problems on the agent of free will.
Else, Sophie wouldn't have much reasons to feel bad for her choice.
>and I would not be to surprised if, upon close inspection and careful definition, it turns out that the agent is identical with the past light cone of the decision.
Perhaps, but the decisions are made at the edge of different light cones -- which are the different persons.
https://plato.stanford.edu/entries/freewill/
http://www.iep.utm.edu/freewill/
>Unfortunately, I was elected to a tutorial fellowship at Merton, my old college in Oxford, and was submerged in the pressures of Oxford tutorial life, so that my Freedom of the Will was not published until 1970. It is not very different from the original article, but does meet some of the criticisms first leveled against it, and sets the Godelian argument in a more general context. Perhaps two points I make in the book are worth reiterating. Professor Minsky spent some time yesterday criticizing Professor Margenau saying that he had misunderstood the import of quantum mechanics. I think actually the misunderstanding was on Professor Minsky's part and that he had failed to understand the structure of Professor Margenau's argument. Quantum mechanics is relevant to the problem of the will because it has replaced classical Newtonian physics which seemed to rule it out. It does not of itself prove free will - lots of quantum mechanical systems have no free will - but it disproves a disproof of it. In my book I referred to speculations on whether quantum mechanics might be replaced by a more determinist theory - a "Hidden Variable Theory" as it is called, and Von Neumann's argument against it. Since then the argument has progressed much further. Von Neumann's proof has been proved and criticized, and a whole series of results have been obtained - Bell's inequalities, Gleason's Theorem, Kochen-Specher Theorem, the two-color theorem: and four years ago Aspect concluded some experiments in Paris, which seemed to rule out any prospect of a hidden variable theory. So it seems to me that Professor Margenau was quite right to see quantum mechanics as bearing on the freewill problem, not as proving that it exists but as, in Plantinga's terminology, defeating a defeater.
further, Lucas proposed the following:
1. Determinism ↔ For any human h there exists at least one (deterministic) logical system L(h) which reliably predicts h's actions in all circumstances.
2. For any logical system L a sufficiently skilled mathematical logician (equipped with a sufficiently powerful computer if necessary) can construct some statements T(L) which are true but unprovable in L. (This follows from Gödel's first theorem.)
3. If a human m is a sufficiently skilful mathematical logician (equipped with a sufficiently powerful computer if necessary) then if m is given L(m), he or she can construct T(L(m)) and Determine that they are true—which L(m) cannot do.
4. Hence L(m) does not reliably predict m's actions in all circumstances.
5. Hence m has free will.
6. It is implausible that the qualitative difference between mathematical logicians and the rest of the population is such that the former have free will and the latter do not.
it is a bit teeth-gnashing though.
The physicist makes some measurements as to the volume of curtain, available air flow, chemical properties of the curtain, flow rate and thermal capacity of the water hose and water, formulating a precise equation for how much water is necessary to put the fire out when starting at a given time, and precisely puts the fire out.
The engineer makes a few quick estimates as the the volume of fuel and air, uses that to estimate the amount of water needed, doubles it, and puts the fire out (with safety factor 2).
The mathematician looks at the fire, looks at the fire hose, states: "The problem is solvable", and exits stage left.
Or, rather, I assert that step 3 posits a degree of self-knowledge that no human actually possesses. Now you have to choose which unproven assertion you're going to believe, his or mine, or else you have to find a way to prove one or the other.
It seems he was ignorant of Bohmian mechanics then, because deterministic QM is still alive and well.
> 3. If a human m is a sufficiently skilful mathematical logician (equipped with a sufficiently powerful computer if necessary) then if m is given L(m), he or she can construct T(L(m)) and Determine that they are true—which L(m) cannot do.
This step is dubious:
1. It's not at all clear that m could prove the truth of T(L(m)) even in these circumstances. I would need to see these steps to be convinced.
2. Humans are finite automotons (see the Bekenstein Bound), and the addition of a "sufficiently powerful computer" drastically increases the computational power available to m beyond what m is capable of alone. So L(m) would have to be a logical system that encompasses this unbounded power, and so T(L(m)) would still have true but unprovable statements, even with the computer due to Goedel.
This argument basically tries to prove that a Turing machine + Oracle can solve the Halting problem for Turing machines, but cannot solve the Halting problem for Turing machines+Oracles. However, it neglects to acknowledge that L is predicated on the Oracle, so we're not solving the Halting problem for Turing machines, we're solving the Halting problem for Turing machines+Oracles given only an Oracle, which is impossible.
As for the argument itself, it's a clever attempt, but still predicated on the assumption that determinism is inconsistent with free will. This has yet to be demonstrated, and in fact, all such arguments to date have been refuted (although clearly not to everyone's satisfaction).
The core of my argument is going to be agency. We are in control because our actions often arise out of our own agency. This as opposed to there being another agent being able to decide our actions.
For this argument to work at all, we need to define agency/being an agent and establish that humans are agents. Moreover, humans must be at least somewhat unique for having agency. Certainly, we don't want rocks, or calculators to be ascribed agency.
I'd say an agent needs 2 things: - Intention - 'Potency', that is the posibility to measure and affect things related to those intentiosn. This has again moved the goalposts, since we now need to establish Intention and potency. I'd posit that, if humans have intention, they certainly have Potency and thus only focus on intention.
Intention arose in humans through evolution. Skipping some steps, at one point organisms started measuring their environment, and responding to it. This is a huge evolutionary advantage (i.e. move towards the food and away from the predators). Simple measurement and response isn't quite enough though, but there is another advantage to be had: modeling. An organism is modeling when it isn't just measuring the outside world, but also making predictions. Once these predictions start taking into account possible actions by that organism, and those predictions start informing the actions of the organism, those actions have 'intent'.
And that is my argument in short form. Free will arises from choosing actions based on their predicted effect in some internal model.
There is some stuff to clear up between being an agent, and having agency over your own actions. It might be some other much more effective agent has more control over you than your own agency. I don't believe any such powerful agents exist though.
Or, a human's agency and the agency of a bad guy in a modern video game with complicated AI?
For the videogame UI, there is a massive limitation on agency because it was fully designed by another agent. That designer agent thus has very large influence over the AI.
I'd allow for complicated AI with emergent behavior to have agency up to a point. As a corollary, this means humans could have agency even if they were created by god.
This is a higher level of self control than a calculator in that it has no capacity to alter its behavior. If a calculator had preferred states, it has no capacity to alter its behavior to achieve those preferred states.
Now, we still don't have the capacity to change our will and so we don't have "ultimate" (contra-causal) free will. But I think its pretty clear that we have more control than a calculator.
I don't mean, harder to predict. I mean, are we not as deterministic as anything else?
The way i see it is that choice/free will is just a useful concept at the level of description of people. The mistake is mixing the levels of description and expecting the resulting statements to be coherent. There's no point in a causal chain where a "choice" happens, yet at the level of description of people with behaviors, choice happens. If we don't expect some kind of metaphysical truth underlying the claim that we make choices, then there's no problem. If you want your choices to have some acausal sort of control, then you're out of luck.
Whenever the question of free will comes up, I remember that phrase and side on the camp of free willers.
Or am I missing some kind of humor there?
It's also a terribly boring problem. Regardless on you stance on what free will is, the answer does not mean you are removed from culpability; it rather means that you are not exempt from defense of your actions as a being with finite knowledge explaining your state to another being (presumably, a court) with finite knowledge.
For all the philosophers asking this question, they fail to produce any instance where the answer is meaningful. Leopold and Loeb managed to persuade a gullible jury that they can excuse human interactions with philosophy; it is of interest to pop psychology and flaws in the american justice system, NOT to philosophy.
The phenomena of consciousness are far more worthwhile concepts because they don't pigeonhole reality into a dry dichotomy.
He's also doing the thing where he's discussing the issue without a direct attempt to resolve it, which may be throwing off your comprehension.
If you acknowledge culpability in a given set of circumstances, then those culpable acted of their own free will. Choice from free will entails moral culpability, so you can only deny free will by also denying moral culpability.
The fact that it's difficult to define the precise meaning of these terms doesn't make the problem meaningless.
All of this smacks of scientists trying to weigh in on ontological problems. Science is only relevant so far as to show us the limits of what we can claim—in this case, science shows us free will doesn't map neatly into our known universe, hence the article above trying to argue for ways to preserve this abstract concept in a world that doesn't care.
Look at: The Emperor's New Mind. It's the best attempt so far at actually fleshing out the problem, and it hinges entirely on the (fallacious) idea that because humans can innovate they aren't bound by turing completeness. Of course, this is hogwash, because what you'd have to demonstrate was that a single consciousness can prove everything given a finite set of axioms, which is of course an impossible proposition. Penrose sidesteps this by focusing on the magnification of quantum effects to human-level.
I mean, come on—if that's your hope at understanding your own brain, just give up and imagine fairies in it. Is that really more comforting (...let alone rational) than accepting you're driven more by your circumstances than by some ineffable thing about yourself you can't even define? Might as well accept that you're inherently an observer to phenomena outside your immediate ability to understand completely—you know, the basis of the scientific theory.
When people talk about free will in terms of quantum effects, all they reveal is how poorly read they are. It's a binary concept that doesn't map at all into a world where things-are—-—of-predictatable-but-you-can't-predict-the-unpredictable. It's nonsensical except at levels that are meaningless to human thought OR meaningless to physics.
I'm sure that's what you believe, but that doesn't make it true.
> do you seriously think that there is an objective culpability for arbitrary concepts when people won't even interpret the arbitrary concepts the same way?
The whole point of objectivity is that interpretation has no bearing on truth (or culpability in this case).
> in this case, science shows us free will doesn't map neatly into our known universe
Science shows no such thing. Assuming determinism of choice shoes only that philosophies of free will that claim free choice is incompatible with determinism are incompatible with evidence. Of course, we don't actually know that we are deterministic (evidence is not conclusive), and it does not in any way rule out free will that is compatible with determinism, which is the type most philosophers profess.
This is an all too common mistake that scientists make: thinking that their definition of free will is obviously the one everyone is using.
There is a part in this documentary that covers a man who had a hypothesis that human beings are driven to do things by their DNA. I'd link the specific scene but I have to go to work, maybe later.
https://arxiv.org/pdf/1611.02269.pdf
What he calls the "free will camp" sounds like a weird mishmash of libertarianism (determinism is false, and we have free will) and hard-determinism (determinism is true, therefore we don't have free will). It completely ignores compatibilism (https://plato.stanford.edu/entries/compatibilism/).
Moreover, the argument he gives regarding Leopold and Loeb is baffling. That a jury has no free will to listen to your argument for hard-determinism is prima-favor irrelevant. A hard-determinist holds that some people are determined to believe in free will, and moral responsibility, but they're incorrect. Any graduate student in philosophy would be expected to note these distinctions.
As for the view that free will is an illusion, he presents one of the canonical arguments uncharitably. Leaving aside the question of whether the invocation of NP actually establishes what he says it does (something non-determined but non-random), the opponent of libertarianism doesn't need the claim that determinism and randomness are our only options. They just need (something like) the premise that if our actions don't follow from our reasons/emotions/volition, they can't be considered free, and that anyone proposing some status between randomness and determinism needs to account for the sense in which decisions flowing from that status are free.
The linked lecture is a solid, engaging and informative discussion of an issue, bringing together many different fields of though, science, and engineering. That sounds like a great thing to have on the front page.
But while the piece was engaging, it wasn't informative, which is precisely my complaint.
Edit: That is, it "taught" me little outside of amazing and novel new perspectives on things I already know.
Near as I can tell, from the Wikipedia explanation, compatabilism is an ethical consideration of free will. Most other philosophy is on a metaphysical level; Scott Aaronson appears to be talking about free will as a physical phenomenon. It make sense that he would't bring up an ethical framework.
Here's my take, which is slightly different in emphasis than the SEP one (which I guess means I am probably wrong/idiosyncratic):
Compatibilism is a metaphysical theory. What is counter-intuitive about the free will debate is that there are two entwined but distinct questions.
The first is (roughly) what is the full physical/causal/predictive/descriptive* story about human decision making. Is it deterministic? Almost deterministic but with some random noise? Something else?
Second, given that description, do we have free will? That is a metaphysical question, though many people are interested because they believe that there are ethical implications (compare: whether God exists is a metaphysical or perhaps empirical question, but it might have some ethical implications).
Compatibilism is primarily an answer to the second question. It says roughly "you might have thought free will means not being determined, but that is an illusion. Actually, free will is completely compatible with determinism. It is something like being able to act on the things that you take to be reasons, even if it's fully deterministic which things you think are reasons."
Talking about free will as a physical phenomenon (which is not quite what I'd describe Aaronson as doing) typically indicates that you've assumed some answer to the second question, so that you're only considering the first.
Btw: one point of view is that the first question makes sense, but the second doesn't. I don't have a developed opinion, as I didn't really work on this topic when I studied philosophy, but I lean that way.
* (none of these terms is perfect, and there's lots of debate about the right characterization here)
(Eek, weird sentence. No awkward grammar fixes welcome.)
However, I find Descartes Meditations gripping for the way he poses the question of whether there can be any solid foundation for knowledge and the resulting skeptical worries (though his solution is somewhat disappointing). Hume's Enquiry Concerning Human Understanding is also wonderful.
If you find free will interesting, Kane's "A Contemporary Introduction to Free Will" is solid--it was the text a professor assigned when I ta'd for his course.
Thanks for the recommendations!
I think the biggest answer that's in here to The Question is this one:
> For me, the glaring fallacy in the argument lies in the implication Not Determined ⇒ Random. If that was correct, then we couldn't have complexity classes like NP---we could only have BPP. The word "random" means something specific: it means you have a probability distribution over the possible choices. In computer science, we're able to talk perfectly coherently about things that are non-deterministic, but not random.
...which, in a way I don't fully grok, seems to be saying there's room between things being predicable (no free will) and random (nonsensical).
Then later, this bit:
> 1) We have the free will to choose in what basis to measure a quantum state. That is, at least the detector settings are not predetermined by the history of the universe. 2) Relativity gives some way for two actors (Alice and Bob) to perform a measurement such that in one reference frame Alice measures first, and in another frame Bob measures first. 3) The universe cannot coordinate the measurement outcomes by sending information faster than light.
which I also still don't really understand but woah.
As a contribution: I encountered an amazing demonstration of "free will", in that it was an experience you could go through, draw a box around, and write "here be free will" on the box.
It's pretty simple: Think of a city, and say it's name aloud.
Now consider what just happened in your mind and brain.
You didn't say the name of a city you've never heard of. You could not do this; you were "not fee" to make that "choice".
You (probably?) though of more than one city, and then selected on. You could do this; you were "free" to make that "choice".
This demonstration doesn't really go after the question as to what you actually have that can be labeled "choice", but it really nicely draws a box for that label to be applied to.
Now if you bring a free will into play, where does it reside? Is it somewhere outside physics? Thats metaphysics then. And out of reach for any understanding anyway so pointless to reason about.
What I could agree on is to see “free will“ as a metaphor for the essence of our existence. The part that we probably never can explain.
Replace 'Free Will' with the most underwhelming term that is for all intents and purposes synonymous for free will in the discussion.
I also liked his book, "The Big Picture".
http://webcache.googleusercontent.com/search?q=cache:gnEjOAn...
Basically, they try to use cosmic "particle selection generators" to try and circumvent the free-will loophole.
There is also a series of recent experiments on using quantum random number generators to circumvent the free-will loophole.